Turing test
Computer intelligence systems are becoming ever more sophisticated, and are able to perform complex, worldly tasks that were previously only available to humans. One example of this is Google’s self driving car. But as systems like these progress, how will we know the current state of artificial intelligence. One potential indicator was put forward by legendary computer scientist Alan Turing.
The Turing Test states that a computer attached to a teletype (a tool of Turing’s era) should be able to conduct a conversation with a human, and the human should not be able to determine whether the agent they are communicating with is a computer or another human. Effectively the computer should be able to conduct a text based conversation with a person as well as a human being could.
The Turing Test became a famous concept in the world of computer science, as an operational definition of intelligence in computers that could easily be measured. The concept was further refined in the field to include the notion of a “total Turing Test,” in which the conversation is conducted over video. Not only would this require the computer to respond to visual cues in the interrogator, but it would require the computer to manipulate physical objects that may be passed “through the hatch” to it.
No comments:
Post a Comment